When a cpufreq-driver wants to create files in sysfs that are already
created by an oher cpufreq-driver it gets -EEXISTS.

Oleksii Kutuzov authored 1 month ago and Tobias Kahlki committed 4 weeks ago

It contains variuos additional stuff for the processor, as some
memory regions, firmware drivers, configs for memory dumps,
thermal zones, dma heaps etc.

Device can be "woken" from suspend/sleep state by pressing SW5 (SLEEP)
button on B79 carrier board

Christos authored 1 month ago and Tobias Kahlki committed 1 month ago

i2c1:
- HT16K33 Auxiliary Display
- TI TMP102 Onboard Temperature Sensor
- TI IN219 Onboard Current Sensor

i2c2:
- LTC4100 Smart Battery Charger Controller
- Atmel 24C32 EEPROM

Christos authored 1 month ago and Tobias Kahlki committed 1 month ago

- LTC4100 Smart Battery Charger Controller
- HT16K33 Auxiliary Display
- Atmel 24C32 EEPROM
- TI TMP102 Onboard Temperature Sensor
- TI IN219 Onboard Current Sensor

During development, we have disabled the device tree nodes for soundwire. While that is in effect, these drivers are not needed.

If the referenced files was to change, the reference we're providing would be out of date. These changes prevent this.

Chris Lew authored 9 months ago and Christos committed 1 month ago

Ignore the ENODEV failures returned by kernel_sendmsg(). These errors
indicate that either the local port has been closed or the remote has
gone down. Neither of these scenarios are fatal and will eventually be
handled through packets that are later queued on the control port.

Signed-off-by: Chris Lew <quic_clew@quicinc.com>
Signed-off-by: Sarannya Sasikumar <quic_sarannya@quicinc.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://lore.kernel.org/r/20240612063156.1377210-1-quic_sarannya@quicinc.com


Signed-off-by: Paolo Abeni <pabeni@redhat.com>

Oleksii Kutuzov authored 1 month ago and Tobias Kahlki committed 1 month ago

We do not have the USB HUB on QPS615.
QPS615 voltages are enabled via the regulators.

Dietmar Muscholik authored 1 month ago and Tobias Kahlki committed 1 month ago

DP adds an i2c bus to the system. If it is not entirely/properly
configured accessing this i2c bus causes a NULL-pointer to be
dereferenced and the system to reboot.

Oleksii Kutuzov authored 1 month ago and Yuri Mazzuoli committed 1 month ago

Added USB_MUX_SEL pin to the firmware revision D8.

USB0 works again in device mode, as well as in host mode.

Backup-switch-mode set to "direct" to automatically switch to battery when power
goes down

Oleksii Kutuzov authored 6 months ago and Yuri Mazzuoli committed 1 month ago

This commit adds the devicetree node "backup-switch-mode" to the rv3028
driver. It supports the following values:
- direct
- level
- disabled

The values correspond to the different BSM options in the datasheet.
When the setting is present in the devicetree, the driver checks the
current mode during the probing phase. If the desired value differs from
the current value, it is written to the RTC and a message is printed to
dmesg, stating the new mode.

Source: seco-ne/kernel/linux-imx-kuk@92daec58

Oleksii Kutuzov authored 1 month ago and Tobias Kahlki committed 1 month ago

Phy interrupts shoild be connected also to the processor,
will be fixed in the next PCB revision.

- correct include guard
- unify regulator configuration a little more
- remove unused stuff
- fix typos in comments

Dietmar Muscholik authored 1 month ago and Tobias Kahlki committed 1 month ago

In the DP-driver there is "qcom,edp-vcc-en-gpio" that a gpio can be
assigned to in the DT and that powers on the display. But this gpio is
switched AFTER the display has been powered on. Thus there is no other
way to turn on a display but using a hog-gpio.

There is no way to assign a backlight-device to (e)DP, not even a PWM,
only a bunch of gpios. So the pwm is configured as gpio, meaning there is
no way to adjust the backlight-brightness but 2 ways to switch the
backlight off, because both gpios must be high.

Dietmar Muscholik authored 1 month ago and Tobias Kahlki committed 1 month ago

This is used to enable an eDP with a hog-gpio

Dietmar Muscholik authored 1 month ago and Tobias Kahlki committed 1 month ago

This also introduces a more logical naming convention:
seco-<soc>-<som>-<baseboard>-<display>.dts

dtsi-files for baseboards are prefixed with "baseboard-" and those for
displays with "display-".

The QCS5430 is a QCS6490 that is lacking some features, so we use the
QCS6490 dtsi-file for both and disable the non-existing features for
QCS5430. The opposite way would be less ugly but would require modification
of the Qualcomm-provided dtsi-files.

Dietmar Muscholik authored 1 month ago and Tobias Kahlki committed 1 month ago

Qualcomm uses off-tree device-tree-overlays for displays. These overlays are
applied to the base device-tree at build time to generate a combined device-
tree.
Because they are maintained in separate repositories they do not necessarily
match the kernel. If so, for the dtc everything seems to be right, because it
has no knowledge about the base device-tree. The errors, like the most common
"unknown label" because of a simple typo, occure later when the overlays are
applied. While the dtc says: this file, this line, this error, the tool that
applies the overlays only says "FDT_ERR_NOTFOUND", hidden somewhere in the
bitbake-logs, and it takes hours and hours of trial-and-error to figure out
what went wrong.
Furthermore, because the overlays are not applied dynamically at runtime but
at build-time, it does not make any sense to use overlays. They cause nothing
but trouble. For this reasons th overlays have been turned to dtsi-files.

Dietmar Muscholik authored 1 month ago and Tobias Kahlki committed 1 month ago

Even without development is unconvenient enough

- Fix the main comments to reflect the devicetree files correctly
- Update the copyright

Rename the B79 include file and also add the file -b79 file extension to
the base devicetrees.